Target Localization Using MIMO-Monopulse: Application on 79 GHz FMCW Automotive Radar

摘要

Automotive radar is widely used for driving safety support and it is a key element of future autonomous vehicles. Radar sensors have the property that the performance is not to be affected by low vision conditions compared with camera sensors or laser based radar sensors, which makes it crucial for the autonomous driving system.Automotive radar system utilizes millimeter-wave band to detect the range, velocity and direction of arrival (DOA). Commercial 24 GHz and 77GHz radar have been well developed in vehicle applications at present, and the future trend is 79 GHz solution with wider 4 GHz bandwidth, which has been defined by the European Commission in 2004 as the frequency allocation for automotive shortrange radar systems. The main modulation method of automotive radar is Frequency-Modulated Continous-Wave(FMCW), which holds advantages including high-resolution distance measurement, quick updating and lower sampling frequency required on hardware.%good functions in the various weather condition.The resolution and accuracy of DOA estimation depend on the number of used antennas and their physical size. To improve the performance advanced algorithms and antenna configurations can be used. So-called super-resolution techniques can improve the estimation performance but suffer when few measurements are available and the number of targets is unknown, and the performance will also be degraded by correlated signals.Monopulse is an established technique for radars for precision angle estimation. It enables DOA realized in only one pulse and requires quite less computation complexity. Conventional monopulse with one transmitter and two receivers and phased array monopulse have been well developed especially in tracking radar system. MIMO-monopulse has been studied in some literature, but most of them utilize distributed MIMO. The research and application of colocated MIMO-monopulse are still in progress. In FMCW automotive radar application, clutter may have a strong effect on DOA detection and estimation of targets with low RCS, e.g. pedestrian. With Space-Time Adaptive Processing(STAP), it is possible to suppress the clutter in the angle-Doppler domain. MIMO-monopulse basically utilizes digital beamformer to generate sum and difference channel, which also allows STAP processing to obtain adaptive sum and difference weighting. With STAP clutter can be canceled, thus false detection of MIMO-monopulse will be avoided. In this master thesis, we propose the solution of angle estimation algorithm using MIMO-monopulse based on an actual automotive radar provided by NXP using colocated MIMO antenna. The algorithm is simulated through Matlab and verified on real experimental data. The performance of the algorithm is compared with conventional DOA algorithm and the advantages and disadvantages are analyzed. A feasible extension to STAP will be discussed to suppress the clutter for preventing false DOA estimation.Since monopulse estimator has been approved high performance in the field of single target tracking (such as low RCS pedestrian), our proposed algorithm is also validated in single and multiple target DOA estimations, which is an attractive scenario for a potential application using MIMO-monopulse.